Managed flood storage option for selected levees along the lower

Contract Report 608

Managed Flood Storage Option for Selected Levees along the Lower Illinois River

for Enhancing Flood Protection, Agriculture, Wetlands, and Recreation

Second Report: Validation of the UNET Model for the Lower Illinois River by

Abiola A. Akanbi and Krishan P. Singh Office of Surface Water Resources: Systems, Information, and GIS

Prepared for the Office of Water Resources Management

March 1997

Illinois State Water Survey Hydrology Division Champaign, Illinois

A Division of the Illinois Department of Natural Resources

Managed Flood Storage Options for Selected Levees

along the Lower Illinois River


Second Report: Validation of the UNET Model for the Lower Illinois River

by Abiola A. Akanbi and Krishan P. Singh Office of Surface Water Resources: Systems, Information, and GIS

Illinois State Water Survey 2204 Griffith Drive

Champaign, Illinois 61820

March 1997

ISSN 0733-3927

This report was printed on recycled and recyclable papers.

CONTENTS

Introduction 1 Hydraulics and Hydrology of the Lower Illinois River.................................................3 Acknowledgments 16

Unsteady Flow Model Structure 17 Method of Solution 21 Boundary Conditions 22 Dendritic River System 23 Lateral Inflow 25 Channel Cross Sections 25 Roughness Coefficient 26 Initial Condition 26

Lower Illinois River Model Setup 27 Model Layout 27 Cross-Sectional Data ...30 Flow and Stage Time-Series Data 35 Channel Roughness 39 Initial Flow Condition 39

Model Calibration and Verification 41 Calibration of 1979 and 1985 Floods 42 Model Verification with Other Floods .63

Peak Stage Reduction from Levee Storage Options 84 Recurrence Interval Flood Profiles 84 Effect of Changing Discharge and Variable Tributary Flows.......................................99 Effect of Levee Flood Storage on Flood Peaks ...........................................................104

Summary 106 Work in Progress 107

References 109

Page

Managed Flood Storage Options for Selected Levees along the Lower Illinois River


Second Report: Validation of the UNET Model for the Lower Illinois River

INTRODUCTION

The primary goal of the Managed Flood Storage Option Project is to evaluate the

benefits of converting at-risk levees and drainage districts to flood storage areas along the

Peoria-Grafton section of the Illinois River. The implementation of this goal will involve

an analysis of historical floods to determine flood magnitudes and frequencies at which

overtopping of levees in this section of the Illinois River occurs. It will also require an

evaluation of the impact and interaction of flow from the tributaries and backwater flow

from the Mississippi River. The different components of the project can be grouped into

two categories: flood frequency analysis and unsteady flow simulations.

The flood frequency analysis involves the development of discharge-frequency

relations for Illinois River gaging station records at Marseilles, Kingston Mines, and

Meredosia, and the gaging stations on the five major tributaries on the Illinois River, using

both log-Pearson Type HI and mixed distributions. Also, stage-frequency relations have

been developed for Illinois River gages at Peoria Lock and Dam (L&D), Kingston Mines,

Havana, Beardstown, La Grange L&D, Meredosia, Valley City, Pearl, Florence, Hardin,

and Grafton. The results of the flood and stage frequency analyses are used to develop

10-, 25-, 50-, and 100-year discharges and stages that are required in the unsteady flow

model simulations. Singh (1996) reported the flood frequency analysis component of the

project. The first phase of the unsteady flow modeling component is presented here.

Modeling of unsteady flow is based on the application of the UNET (HEC, 1993)

hydrodynamic model originally developed by the U.S. Army Corps of Engineers

(USCOE). The purpose of the unsteady flow modeling is to compute water surface

profiles for the 10-, 25-, 50-, and 100-year floods using the hydrographs obtained from the

1

stage-discharge frequency analyses (Singh, 1996) as boundary conditions. The water

surface profile computations and model verification constitute the first phase of the

unsteady flow modeling component and are reported herein. A third project report will

include subsequent analyses of several levees in the La Grange Pool and a few in the Alton

Pool for flood storage based on the current flood protection levels, maintenance costs and

problems, and net agricultural benefits and economic viability. Each of the selected levees

will be represented in the UNET model with an opening at the crown (500- to 1000-feet

wide and 4- to 6-feet deep) so that floodwaters can flow into the levee district for

temporary storage when the river stage exceeds the stage corresponding to about a 20-

year flood. Water surface profiles will be generated for various combinations of candidate

levees and cut dimensions. The simulation results will be used to develop guidelines for

width and depth of cuts and to select those at-risk levees most suitable for conversion to

managed flood storage options.

Since most of the levee districts along the Illinois River are below Peoria, the

lower Illinois River from Peoria Lock and Dam (L&D) to Grafton was selected as the

study reach Because of the pools created by the La Grange L&D in the Illinois River and

Lock and Dam 26 in the Mississippi River at Alton, the study reach was divided into two

segments: the upper reach from Peoria L&D to La Grange L&D (River Mile 157.85 to

80.1), and the lower reach from La Grange L&D to the Illinois River mouth at Grafton

(River Mile 80.2 to 0). The UNET model was therefore set up as a system of two

separate reaches (La Grange and Alton Pools) and as a single reach from Peoria to

Grafton.

In order to apply the UNET model to predict floods in the lower Illinois River, the

model parameters have to be calibrated to several historical flood events. In the analysis

reported herein, the UNET model was calibrated by matching computed flood

hydrographs at the location of the gages on the Illinois River with the observed flood

records for 1979 and 1985. The accuracy of the calibrated parameters was then verified

by simulating the 1973, 1974, 1982, and 1983 floods. As previously discussed, the next

steps in the unsteady flow modeling involve the prediction of the 100-year and other

frequency flood elevation profiles in the La Grange and Alton Pools, the reduction in the

2

frequency and magnitude of flooding that would result from the impacts of the flood

storage options, and the simulation of several scenarios involving combinations of at-risk

levees and tributary inflow timings in order to evaluate the impact of the storage options

on peak stages.

Hydraulics and Hydrology of the Lower Illinois River

The section of the Illinois River that is under consideration starts from River Mile

(R.M.) 157.7 at Peoria to the confluence with the Mississippi River at Grafton. It is

designated as the lower Illinois River. It is, for the most part, a winding waterway with

wide, flat floodplains and steep bluffs on one or both sides. The channel varies from about

450 feet to 1100 feet in the La Grange Pool, and the valley width varies from about 2

miles near Pekin to about 5 miles near Beardstown. Downstream of Beardstown, the

channel and valley are wider and there are more lateral lakes and ponds. The USCOE

maintains a 300-foot-wide navigational channel.

Five locks and dams are operated on the Illinois River to maintain a 9-foot depth

of channel for navigational purposes. Two of these locks and dams (Peoria and La

Grange) fall within our study area (Figure 1). The only other lock and dam (L&D) that

affects the lower Illinois River is L&D 26 on the Mississippi River, and 15 miles

downstream from the mouth of the Illinois River. Peoria L&D and La Grange L&D were

built between 1936 and 1939. The two structures have similar configurations and were

designed with wicket gates that can be adjusted to maintain the 9-foot navigational depth.

The wicket gates are lowered during high flows to create an open river condition. Both

stuctures were modified between 1986 and 1990 by replacing some of the wicket gates

with taintor gates. The wicket gates are now raised to full height during low flows and the

water depth is controlled with the taintor gate. Figure 2 depicts the taintor gate structure

and the wicket dam at Peoria. Figure 3 depicts the La Grange L&D.

The Peoria lock is located on the east bank of the Illinois River and is 100 feet

wide and 600 feet long. Next to the lock is a taintor gate with 76-foot opening and a dam

436 feet long comprising 116 wickets that are 3.75 feet wide and 16.42 feet high. A

concrete-regulating dam 34 feet long and having an abutment with a single inlet and six

3

Figure 1. Drainage area of the lower Illinois River (shaded), major tributaries, gaging stations, La Grange Pool (Peoria Lock and Dam to La Grange Lock and Dam), and Alton Pool

(La Grange Lock and Dam to confluence with the Mississippi River)

4

Figure 2. Aerial view of the Peoria Lock and Dam during flood stages when the wicket gates are lowered: (a) looking upstream, the lock and adjacent taintor gate are

at the center of the photo and the Peoria Bridge is near the top of the photo and (b) looking downstream from Peoria Bridge

5

Figure 3. Aerial view of the La Grange Lock and Dam under flooding conditions: (a) looking upstream and (b) looking downstream, the lock and taintor gate are

on the west bank and the undersluice gates are impounded

7

outlet ports containing butterfly valves is located west of the wicket gates. An earth dike

extends from the regulating dam to the west bank.

The La Grange lock on the west bank of the Illinois River has the same dimensions

as the Peoria lock. The L&D has a taintor gate and a dam that consists of wicket gates

similar to the ones at Peoria. The dam is 436 feet long from the taintor gate to the

regulating dam on the west bank. The concrete regulating dam is 136 feet long and has 12

equally spaced butterfly valves 6 feet × 6 feet. A 390-foot earth dam connects the

regulating dam to the west bank of the river.

Figure 4 shows the levee and drainage districts (LDDs) located downstream of

Peoria L&D. There are 13 active LDDs in each of the La Grange and Alton pools.

Table 1 contains the list of the LDDs, the riverbank on which each levee is located, the

riverfront extent of the levee, the area within each LDD, the year the district was

organized, and other pertinent information.

In the upper reach of the study area, the four major tributaries that drain into the

La Grange Pool are the Mackinaw, Spoon, Sangamon, and La Moine Rivers (Figure 1).

In the Alton Pool, the only major tributary is Macoupin Creek. The drainage area of the

segment of the Illinois River watershed that drains into La Grange Pool is 11,094 square

miles (sq mi), and the watershed area for the Alton Pool up to the Mississippi River

confluence is 3,258 sq mi. The combined drainage areas of the five major tributaries

constitute about 90 percent of the total drainage area. The drainage area of the Sangamon

River watershed alone is about 50 percent of the entire lower Illinois River watershed area

(below Peoria L&D). Table 2 shows the drainage areas of some of the tributaries

(available from the existing records) flowing into the lower Illinois River.

There are several gages located on the drainage area of the lower Illinois River

(Figure 1). On the Illinois River, there are five gages in the La Grange Pool and six gages

in the Alton Pool. Some of the gages are maintained by the U.S. Geological Survey

(USGS) and the stage-only stations are under the jurisdiction of the USCOE. Table 3 lists

the river mile locations of the gaging stations, the corresponding upstream drainage areas,

gage elevations, and maximum flood elevations during the specified periods of records.

Only the records after 1940 are used in the analyses.

9

Figure 4. Levee and Drainage Districts in the La Grange and Alton Pools of the lower Illinois River

10

Table 1. Active Drainage and Levee Districts along the Illinois River

Approximate Design riverfront Drainage crown

Bank limits of district elevation Levee district position levee (mile) (acreage) (feet-msl) Year organized Remarks

Peoria County Pekin & Lamarsh Right 155.3-149.7 2,722 458.7-458.1 1888

Tazewell County Spring Lake Left 147.5-134.0 13,100 459.0-456.0 1903

Fulton County Banner Special Right 145.5-138.0 3,957 457.0-456.0 1910 Also Peoria County East Liverpool Right 131.7-128.5 2,885 455.4-455.0 1916-23 Liverpool Right 127.0-126.3 2,885 455.0 1916-23 Thompson Lake Right 126.0-121.0 5,498 453.0-451.0 1918-21 Lacey, Langellier, W. Matanzas &

Kerton Valley Right 119.5-112.0 7,800 455.3-454.6 1893,1913,1917

Schuyler County Big Lake Right 108.3-102.8 3,401 451.7-451.2 1905 Kelly Lake Right 102.6-100.4 1,045 456.0-455.0 1916 Coal Creek Right 92.0-85.0 6,400 454.3 1895 Crane Creek Right 85.0-83.6 5,417 451.0-450.0 1908

Cass County S. Beardstown & Valley City Left 87.5-79.0 11,600 455.2-454.4 1913 Also Brown County Meredosia Lake & Willow Left 78.0-73.0 8,089 449.0 1893,1904 Also Morgan County

Table 1. Active Drainage and Levee Districts along the Illinois River (Concluded) Approximate Design riverfront Drainage crown

Bank limits of levee district elevation Year Levee district position (mile) (acreage) (feet-msl) organized Remarks

Brown County Little Creek Right 78.8-75.2 1,611 1893 McGee Creek Right 75.2-67.3 10,800 444.0 1905 Also Pike County Coon Run Left 72.7-67.1 4,511 1902 Also Morgan County

Pike County Valley City Right 66.3-63.0 4,750 1920

Scott County Mauvaise Terre Left 67.0-63.0 3,961 1902 Includes Roberson

Private Levee District Scott County Left 63.0-56.8 10,245 446.0 1909 Big Swan Left 56.7-50.2 12,055 443.0 1903

Greene County Hillview Left 50.0-43.2 12,396 443.0 1906 Also Scott County Hartwell Left 43.0-38.3 8,696 441.0 1906 Keach Left 38.0-32.7 8,000 440.0 1922 Eldred & Spanky Left 32.2-23.6 9,300 440.0-436.0 1909,1917

Jersey County Nutwood Left 23.3-15.0 10,638 438.0 1907

Note: Data were abstracted from the U.S. Army Corps of Engineers Publications USCOE (1987) and USCOE (1996).

Table 2. Illinois River Tributaries below Peoria Lock and Dam

Tributary Illinois River mile Drainage area (sq mi)

Otter Creek (L) 14.7 89.9 Macoupin Creek (L) 23.2 961.0

Panther Creek (R) 36.7 - Apple Creek (L) 38.3 406.0 Hurricane Creek (L) 43.2 - Sandy Creek (L) 50.0 166.0 Little Blue Creek (R) 54.1 - Walnut Creek (L) 56.7 - Big Blue Creek (R) 58.2 - Mauvaise Terre Creek (L) 63.2 178.0 Coon Run (L) 66.9 - McGee Creek (R) 67.0 444.0 Willow Creek (L) 71.2 - Camp Creek (R) 75.7 - Indian Creek (L) 78.7 286.0 Little Creek (R) 78.8 -

La Moine River (R) 83.5 1,350.0 Crane Creek (R) 84.9 40.2 Lost Creek (R) 89.0 16.5 Sugar Creek (R) 94.2 162.0

Sangamon River (L) 98.0 5,419.0 Elm Creek (R) 102.7 9.2 Wilson Creek (R) 108.3 13.3 Otter Creek (R) 111.8 126.0

Table 2. Illinois River Tributaries below Peoria Lock and Dam (Concluded)

Tributary Illinois River mile Drainage area (sq mi)

Spoon River (R) 120.5 1,855.0 Quiver Creek (L) 122.6 261.0 Big Sister Creek (R) 126.3 28.4 Buckheart Creek (R) 128.2 20.6 Duck Creek (R) 131.7 20.5 Copperas Creek (R) 137.4 127.0 Little LaMarsh Creek (R) 147.2 -

Mackinaw River (L) 147.7 1,136.0 LaMarsh Creek (R) 149.7 40.2 Lost Creek (L) 151.0 23.0 Lick Creek (L) 156.4 19.2

Note: Data were abstracted from Healy (1979) and USCOE (1987). L = Tributary is on the left bank looking downstream; R = Tributary is on the right bank looking downstream.

Table 3. Gaging Stations on the Illinois River and Its Major Tributaries Distance

from Record max. daily river Drainage water surface Zero gage

USCOE USGS mouth area elevation elev. Period of Gage station location station # Station # (mile) (sqmi) (feet-msl and date) (feet-msl) record

Mississippi R. at Grafton 0218A 05587450 -0.2 28,906 441.80 (8/1/93) 403.79 1879-1993 Illinois R. at Hardin IH21 05587060 21.5 28,690 442.30 (8/3/93) 400.00 1932-1993 Illinois R. at Pearl IP43 05586450 43.2 27,136 442.75 (8/3/93) 1878-1993 Illinois R. at Florence IF56 56.0 26,919 443.60 (8/1/93) 1930-1993 Illinois R. at Valley City IVC61 05586100 61.3 26,564 444.91 (5/26/43) 414.10 1938-1993 Illinois R. at Meredosia IM70 05585500 71.3 26,028 446.69 (5/26/43) 418.00 1938-1993 Illinois R. at La Grange L&D (TW) 05585300 80.2 25,648 447.10 (5/26/43) 406.00 1937-1993 Illinois R. at Beardstown 11-0492-6 05584000 88.8 24,227 449.50 (5/26/43) 419.89 1940-1993 Illinois R at Havana 11-3940-4 05570520 119.6 18,299 451.00 (5/26/43) 424.40 1878-1993 Illinois R at Kingston Mines 05568500 144.4 15,819 454.00 (5/25/43) 428.00 1940-1993 Illinois R. at Peoria L&D (TW) 05560000 157.8 14,554 455.90 (5/24/43) 417.00 1940-1993 Spoon River at Seville 05570000 38.7 1,636 467.04 1914-1993 La Moine River at Ripley 05585000 12.3 1,293 431.10 1921-1993 Macoupin River at Kane 05587000 16.1 868 426.77 1921-1993 Mackinaw River near Green Valley 05568000 13.7 1,089 477.11 1921-1993 Sangamon River near Oakford 05583000 25.7 5,093 452.88 1910-1993

Notes: TW = Tailwater USGS = U.S. Geological Survey L&D = Lock and Dam sq mi = square mile USCOE = U.S. Army Corps of Engineers msl = mean sea level

Acknowledgments This study is jointly supported by the Office of Water Resource Management,

Illinois Department of Natural Resources, and the Illinois State Water Survey. Gary Clark, Office of Water Resource Management, is serving in a liaison capacity during the entire course of the study. The U.S. Army Corps of Engineers District office in Rock Island, Illinois, provided some of the basic data used in this study. Computer runs for unsteady flow simulations were carried out by graduate research assistant Miguel Restrepo, who is studying for his Ph.D. in Natural Resources and Environmental Sciences at the University of Illinois. Norma Lee Rhines typed the final manuscript, Linda Hascall assisted with the graphics, and Eva Kingston edited the report.

16

UNSTEADY FLOW MODEL STRUCTURE

To simulate the flow of water in a river channel, differential equations of mass and

momentum are solved with a numerical method. Depending on the application, the partial

differential equations could be used to describe steady and unsteady flow situations.

Steady-state flow is based on the assumption that flow depth and discharge do not vary

with time. It is well known that the flow in most rivers is unsteady. However, if the

variation in the average discharge for several days following a storm and prior to the next

storm is small, the flow can be considered steady flow for all practical purposes.

Due to the complexity in the numerical solution of the three-dimensional flow that

is actually occurring in a river and the exorbitant cost of obtaining adequate field data,

certain assumptions can be made to reduce the dimensions and hence the complexity of the

hydrodynamic model. For instance, by assuming hydrostatic pressure conditions, the

three-dimensional hydrodynamic equations will be reduced to those for a two-dimensional,

vertically integrated model. Moreover, if the principal flow direction can be assumed to

follow the centerline of the channel, the equations are further simplified and a one-

dimensional set of equations is obtained. The latter assumption implies that the lateral

variations in water depths and velocities are negligible. Examples of one-dimensional

unsteady flow models are FLDWAV (Fread, 1985), FEQ (Franz, 1990), and UNET

(HEC, 1993). These models are widely used for large river systems such as the

Mississippi and Chesapeake Bay. Two- and three-dimensional models are mainly applied

to short channel reaches to examine flow conditions around bridge piers, bends, and

channel confluence. They have also been used to simulate tidal flow in lagoons and

estuaries. Examples of two-dimensional models include FESWMS-2DH (Froehlich, 1989)

and TABS-2 (Boss International, 1993).

The equations of conservation of mass and momentum for one-dimensional flow

are based on the Saint-Venant derivations. The assumptions used in the derivation of

these equations are:

• Constant average velocity at any section is perpendicular to the main flow direction.

17

• Negligible vertical acceleration implies that hydrostatic pressure predominates in the

fluid, which means that the slope of the water surface profile varies gradually.

• Main flow direction along the stream centerline can be approximated with a straight

line.

• Channel bottom slope is very small such that tan Ψ = sin Ψ, where Ψ is the angle of the

channel bed to the horizontal surface.

• No scouring or deposition of sediment occurs on the channel bed.

• Channel roughness for steady flow is applicable for unsteady flow.

• Fluid is incompressible and has homogeneous density.

The mass and momentum equations can be expressed as:

(1)

and

(2)

where x is the distance along the channel, t is the time, Q is the flow, A is the cross-

sectional area, h is the water depth, S is the storage volume per unit length in the direction

of flow, Sf is the frictional slope, v1 is the lateral inflow velocity, g is the gravitational

acceleration, and q1 is the lateral inflow per unit distance.

Most rivers during flood flow usually convey water within the channel banks and

in the floodplains. The flow is essentially two-dimensional flow if depth averaging is

assumed. The mass and momentum equations can be written for both the channel and

floodplains as:

18

Channel:

(3)

(4)

Floodplain

(5)

(6)

Mc and Mf are the momentum flux exchanges per unit distance between the channel and

floodplain, respectively; is the fraction of the momentum entering the receiving stream;

and the subscripts c and f represent the channel and floodplain. This momentum flux is the

momentum of the flow passing through the channel section per unit time per unit distance

along the channel. The water surface elevation is assumed to be the same for the channel

and floodplain. Since the exchanges of mass between the channel and floodplain are

equal, then qc ∆xc = qf ∆xf, where ∆xc and ∆xf are the lengths of the shoreline and bluff

across which lateral inflow enters the channel and floodplain, respectively. Equations (3)

and (5) can be manipulated to yield

(7)

19

where is equal to ∆xc / ∆xf and Q = Qc + Qf.

Since the momentum exchanges between the channel and floodplain flows are also

equal, i.e., Mc ∆xc = -Mf ∆xf , then equations (4) and (6) can be combined to yield the

following expression:

(8)

If an equivalent frictional force is defined as

(9)

and a velocity distribution factor, β, as

(10)

then equation (8) can be expressed in simplified form as

(11)

where ∆xe is the equivalent flow path, Sf is the frictional slope for the entire cross section,

and A =Ac + Af is the total cross-sectional area.

20

Method of Solution

Equations (7) and (11) can be solved by finite difference or finite element methods.

The finite element method is usually very cumbersome and may produce unstable solutions

for hyperbolic equations unless upwinding terms are added to the numerical scheme

(Akanbi and Katopodes, 1988). Finite difference methods, on the other hand, are easier to

develop and have been widely applied to fluid flow problems. Explicit finite difference

schemes are used to solve for the unknown nodal variables at a new time step by using the

previous time step solutions. However, for implicit schemes, the nodal solutions are

solved simultaneously for all nodes at the new time step. An initial guess of the solution is

made at the beginning of a new time step, and an iterative procedure is then used to

update the variables until a specified error criterion is satisfied. Examples of finite

difference schemes that have been applied to fluid flow problems include the Method of

Characteristics, Leap Frog, Lax-Wendroff, and Four-Point or Box schemes.

Equations (7) and (11) have been solved with the implicit four-point scheme. The

weighted four-point implicit scheme was first developed by Preissman (1960). The

discretization of the depth of flow, h, and its derivatives are given by:

(12)

(13)

(14)

where is the weighting factor, which varies between 0 and 1. When 9 is zero, an explicit

scheme is obtained. The scheme is fully implicit when 9 is equal to 1. When the

derivatives in equations (7) and (11) are replaced with the finite difference schemes

21

(equations 12-14), and by grouping the unknown quantities on the left-hand side of the

resulting expressions, the following linear algebraic equations are obtained:

(15)

(16)

The matrix coefficients Aj,Bj,Cj,Dj,Ej,Aj,Bj,Cj,Dj, and Ej are defined in the

appendix. ∆h and ∆Q are the changes in h and Q between two consecutive time steps.

Equations (15) and (16) are combined to form a single matrix of equations which is solved

with the Gaussian Elimination Procedure (Bathe and Wilson, 1976; HEC, 1993). The

Gaussian Elimination technique converts the sparse matrix into a triangular system which

is easier to solve.

Boundary Conditions

In the UNET model, a simple reach of a river will be subdivided into N-l finite

difference cells, which are bordered by N computational nodes. By writing continuity and

momentum equations for each cell, a total of 2N-2 equations will be formed. However,

since there are 2N unknown h and Q at the N nodes, two additional equations are

needed in order to determine all unknown quantities. These extra equations are provided

by the boundary conditions at the upstream and downstream ends of the reach. The

UNET model simulates subcritical flow, the prevalent condition in most natural streams.

However, the model allows the user to specify supercritical flow condition at the upstream

boundary.

The upstream boundary condition can be specified as a stage or water surface

elevation hydrograph; or as a discharge hydrograph. The downstream boundary condition

can be a stage hydrograph, a discharge hydrograph, or a known relation between stage

and discharge such as a rating curve. The specification of any one of these boundary

22

∆ ∆

conditions will be satisfactory. If the stage is used, then the discharge, Q, is computed in

the solution of the Saint-Venant equations. Similarly, if the discharge is specified, then the

stage, h, is computed.

Any combination of upstream-downstream boundary conditions can be specified.

It has been observed that when discharge hydrographs are prescribed for both upstream

and downstream boundary conditions, any error in the initial conditions will be magnified

as the solution progresses. However, the errors are usually damped out after a few time

steps for the other combinations of boundary conditions.

In a typical application of the UNET model for flood flow prediction, the upstream

boundary condition will usually be a discharge or stage hydrograph while at the

downstream boundary a stage or stage-discharge rating relation is specified. A stage

hydrograph is usually prescribed at the downstream boundary when the boundary is

influenced by tidal actions.

Dendritic River System

In a dendritic river system, the main stem is subdivided into several subreaches

such that each of the subreaches represents a segment of the river between the confluence

of two adjacent tributaries. For a tributary that experiences significant backwaters due to

floods on the main river and for which cross-sectional data are available, the tributary can

be divided into several cells, the total number of cells depending on the degree of

variability of the cross-sectional geometry along the river. For tributaries without

available cross-sectional geometry data or not affected by backwater flow, the tributary

flow can be input into the model as a point or a uniformly distributed inflow along the

banks of the main river stem.

Figure 5 is an example of a river system that has two tributaries. The main river

stem is divided into three reaches comprising the river segment between the upstream end

and the mouth of tributary #1, the river segment between the two tributaries, and the river

segment from the mouth of tributary #2 to the downstream end. There will be a total of

five reaches if the two tributaries are included. Each of the reaches has been divided into

finite difference cells bounded by nodes. The nodes correspond to the locations where

23

Figure 5. A two-tributary river system is divided into subreaches and finite difference cells

24

cross-sectional geometries have been measured or estimated. The nodes are numbered

starting at the upstream end of a reach and increase sequentially towards the downstream

end. The node numbering commences at the upstream end of reach 1, the uppermost

reach of the main stem, to the downstream end of the reach. The numbering continues

from the downstream end of reach 1 to the upstream boundary of reach 2 (tributary #1).

This node numbering procedure is repeated for reaches 3, 4, and 5. The node numbers

and the total number of cells for the five reaches in this example are given below:

Node Numbers

Reach Upstream/Downstream Total Number of Cells

1 1-9 8

2 10-18 8

3 19-25 6

4 26-30 4

5 31-34 3

Lateral Inflow

UNET incorporates tributary inflows through the lateral inflow term, ql, in

equations (7) and (11). ql is specified as either a point inflow hydrograph or a uniformly

distributed hydrograph between any two specified cross sections. For point inflow, ql will

represent the total flow entering the reach. The effect of the lateral inflow will be

observed at the next downstream cross section.

Channel Cross Sections

A channel cross section can be represented by regular or irregular geometry. The

cross sections are usually taken at gaging station locations, at locations where the changes

in cross-sectional geometry significantly affect the flow, at the confluence of tributaries,

and around hydraulic structures such as bridge piers, culverts, weirs, and locks and dams.

The surveyed sections are input into the model as pairs of elevation and distance from a

predetermined point on the left bank of the stream looking in the downstream direction.

25

The UNET model also allows the demarcation of active flow areas as well as dead storage

areas in a cross section. This enables the model to adequately represent bank

encroachment, embankments, bridge piers and openings, levees, and floodways.

Roughness Coefficient

The Manning's roughness coefficient, n, is used to describe the resistance to flow

due to bed forms, vegetation, bends, and eddies. Roughness coefficients are specified for

the bank-full channel area, and for the left and right overbank areas of each cross section.

The model computes area-weighted average roughness values for the left and right

overbank areas in order to determine the roughness value for the floodplain component of

the flow. Also, the model allows conveyance, which is inversely proportional to the

Manning's roughness, to be varied with discharge.

Initial Condition

Initial flow distributions are required for each of the reaches. For instance, if the

initial flow condition is assumed to be steady state in the example in Figure 5, five flow

distributions will be specified for the river system. Required model input also includes the

initial water surface elevation for each of the storage areas if there are any in the river

system.

26

LOWER ILLINOIS RIVER MODEL SETUP

The UNET model can simulate one-dimensional flow through single, dendritic, or

looped systems of open channels. The model can simulate the interaction between channel

and floodplain flows; levee failure and storage interactions; and flow through navigation

dams, gated spillways, weir overflow structures, bridges and culverts, and pumped

diversions.

The model solves the one-dimensional unsteady flow equations using a linearized

implicit finite difference scheme. It requires a stage or discharge condition at the upstream

boundaries of the main river and tributaries which have existing cross-sectional data. A

stage or stage-discharge relation is prescribed at the downstream boundary of the main

river. Other input requirements include tributary inflows, cross-sectional geometry, and

hydraulic roughness parameters. The cross-sectional geometry and the stage or discharge

boundary condition are prepared in two separate files. In addition, the model can read

time-series data of stage and discharge from the Data Storage System (DSS) database that

developed by the USCOE for the Hydraulic Engineering Center's (HEC) series of models.

The output of the UNET model includes the time-series of stage and discharge at

prescribed locations and plots of water surface elevation profiles. Figure 6 shows a

flowchart representing the components of the UNET model and its relations with the DSS

package.

Model Layout

As discussed in the previous chapter, the river system is usually set up as a series

of interconnected reaches based on the number of tributaries and the location of their

confluences with the main river. In the case of the lower reach of the Illinois River (Peoria

Lock and Dam to Grafton), five gaged tributaries contribute significant flow to the Illinois

River. Other tributaries to the Illinois River are ungaged. The Sangamon River is the only

tributary that has existing cross-sectional data. Because of the lack of such data on the

other major tributaries, the lower Illinois River has been set up as a system of three river

reaches as shown in Figure 7. In this figure, Reach 1 is the segment of the Illinois River

27

Figure 6. A flowchart represents components of the UNET model and its linkage with the Data Storage System database

28

Figure 7. This model represents the lower Illinois River as a system of three river reaches and several lateral inflow areas

29

from Peoria Dam to the river section immediately upstream of the Sangamon River

junction. Reach 2 is the segment of the Sangamon River from the gage at Oakford to the

river mouth, and Reach 3 is the segment of the Illinois River from the Sangamon River

junction to the Illinois River mouth at Grafton.

The Mackinaw and Spoon Rivers are the major tributaries in Reach 1 while the

La Moine River and Macoupin Creek are the major tributaries in Reach 3. These

tributaries and the smaller ones are taken as lateral inflow at their confluence with the

Illinois River. The Sangamon River (downstream of Oakford) was assumed to have no

point lateral inflow from its tributaries. Instead, the contributions from its tributaries were

assumed to be uniformly distributed lateral inflow along the entire length of the Sangamon

River. Figure 7 also shows the location of the discharge/stage gages on the Illinois River

at Peoria L&D, Kingston Mines, Havana and Beardstown in Reach 1; and La Grange

L&D, Meredosia, Valley City, Florence, Pearl, Hardin, and Grafton in Reach 3. It will be

shown in the model calibration section below how stage and discharge hydrographs,

generated by the model at these gage locations, are matched with historical records.

Cross-Sectional Data

The USCOE provided 412 surveyed cross sections from for the Peoria-Grafton

section of the Illinois River. Thirty-three surveyed cross-sections were also obtained from

the USCOE for the Sangamon River, starting from the gage at Oakford and extending to

its confluence with the Illinois River. Figure 8 shows some typical cross sections at

selected locations on the Illinois River.

Hydraulic structures such as levees, weirs, locks and dams, bridge piers,

embankments, channel encroachments, and storage areas are given special considerations

in the model. Within the selected study reach, the only major hydraulic structure, apart

from bridge piers, is the La Grange L&D at river mile 80.2. The L&D has a dam section

comprising wicket gates, a taintor gate structure, a regulating dam with butterfly valves,

and a 390-foot earth dam.

There are 10 levees and drainage districts in Reach 1 and 15 LDDs in Reach 3.

Table 4 lists data on the drainage area behind the levees, the average round elevation in

30

Figure 8. Selected cross sections of the lower Illinois River in the Alton Pool at River Miles (a) 8.7 and (b) 67.8, and in the La Grange Pool at River Miles (c) 81.6 and (d) 132.2

31

Figure 8. (concluded)

32

Table 4. Model Input Parameters for Levees and Drainage Districts

Cross section

Protected Interior Breach upstream Crown elevation area elevation River Mile elevation of levee Upstream Downstream

Levee district (acres) (feet-msl) Upstream Downstream (feet-msl) (mile) (feet-msl) (feet-msl)

Peoria L&D to Sangamon River Pekin& La Marsh 2,722 438.0 155.3 149.7 458.0 155.60 458.7 458.1 Spring Lake 13,100 430.0 147.5 134.0 455.0 151.20 459.0 456.0 Banner Special 3,957 440.0 145.5 138.0 455.6 145.70 457.0 456.0 East Liverpool 2,885 435.0 131.7 128.5 455.0 132.20 455.4 455.0 Liverpool 2,885 430.0 128.1 126.0 455.0 128.40 455.0 455.0 Thompson Lake 5,498 430.0 126.0 121.0 453.0 126.40 453.0 451.0 Lacey, Langellier, W. Mantaza & 7,800 435.0 119.5 112.0 455.0 119.56 455.3 454.6

Kerton Valley Big Lake 3,401 435.0 108.3 102.8 451.0 108.40 451.7 451.2 Kelly Lake 1,045 434.0 102.6 100.4 455.0 102.70 456.0 455.0

Sangamon River to Grafton Coal Creek 6,400 430.0 92.0 85.0 454.7 92.20 454.3 454.3 Crane Creek 5,417 430.0 85.0 83.6 450.0 85.50 451.0 450.0 S. Beardstown & Valley City 10,516 428.0 88.2 79.0 453.8 88.40 455.2 454.4 Little Creek 1,610 426.0 78.2 75.1 448.0 78.50 448.0 448.0 McGee Creek 12,400 430.0 75.0 67.1 445.5 75.50 445.5 445.5 Coon Run 6,162 438.0 72.7 67.1 448,5 72.80 448.3 448.5 Mauvaise Terre 6,626 440.0 65.8 63.4 446.0 66.00 446.0 446.0 Valley City 4,700 435.9 66.2 62.5 446.0 66.60 446.0 446.0 Scott County 12,700 433.6 63.1 56.7 446.0 63.30 446.0 446.0

Table 4. Model Input Parameters for Levee and Drainage Districts (Concluded)

Cross section

Protected Interior Breach upstream Crown elevation area elevation River Mile elevation of levee Upstream Downstream

Levee district (acres) (feet-msl) Upstream Downstream (feet-msl) (mile) (feet-msl) (feet-msl)

Big Swan 14,200 435.0 56.5 50.1 444.0 57.00 444.0 444.0 Hillview 13,700 427.9 50.0 43.2 443.5 50.05 443.5 443.5 Hartwell 9,300 426.9 43.1 38.2 442.5 43.17 442.5 442.5 Keach 9,700 429.9 38.0 32.8 441.5 38.20 441.5 441.5 Eldred & Spanky 9,800 427.4 32.4 23.8 441.5 32.70 441.5 441.5 Nutwood 10,600 428.3 23.6 15.1 440.0 24.26 440.0 440.0

Note: Data were abstracted from USCOE (1987, 1996).

the levee districts, upstream and downstream riverfront limits, and upstream and

downstream crown elevations for Reaches 1 (Peoria-La Grange) and 3 (La Grange-

Grafton).

The data input file includes cross sections of bridge crossings at river miles 71.31

(Main Street Bridge at Meredosia), 61.39 (Norfolk & Western Railway), 55.96 (U.S.

Highways 36 and 54 or Florence Highway), and 21.65 (State Highway 100). However,

the model did not include other bridge crossings at river miles 152.9 (Pekin Highway

bridge), 152.3 (Chicago & Northwestern Railway), 119.5 (U.S. 136), 88.8 (Burlington

Northern Railroad), 87.9 (Beardstown Highway or US 67 and State Route 100) and 43.2

(Illinois Central & Gulf Railroad or Alton Route) because the spans between adjacent

piers were wide enough such that they do not significantly affect the river flow.

Flow and Stage Time-Series Data

The ability of the UNET model to use either the discharge or the stage upstream

boundary condition, as previously discussed, is advantageous in situations where the

upstream end of the reach is a stage-only station. This is the case at Peoria L&D where

only stage records are available (1940 -1993) from the USCOE. The only other upstream

boundary in the system is on the Sangamon River (Reach 2) at the site of the Oakford

gage. This station has a long record of discharge (1910-1993). The downstream

boundary is at Grafton (Reach 3), and the prescribed boundary condition is a time-series

of stage for the model calibration and verification. A stage-discharge rating relation is used

to simulate levee storage options.

A time-series of discharges is required for point lateral inflows from some of the

tributaries listed in Table 2. The gaged tributaries have already been identified as the

Mackinaw, Spoon, Sangamon, and La Moine Rivers, and Macoupin Creek. The lateral

inflows from ungaged tributaries with drainage areas greater than 50 sq mi were assumed

to be point discharges to the Illinois River. However, for all ungaged tributaries, within a

particular segment, with drainage areas less than 50 sq mi, the flows were combined and

assumed to be uniformly distributed as lateral inflows along that segment of the Illinois

River.

35

Ungaged tributary flows can be estimated in several ways. For the present

analysis, the ungaged tributary flows were estimated using records from a gage in a nearby

watershed with similar hydrologic pattern. The discharge record at the selected gage will

be scaled with the fraction of the area of an ungaged watershed to the area of a gaged

watershed. Table 5 shows the river mile location, drainage area, drainage area ratio, and

time lag for the ungaged tributaries and drainage areas. The time lag was estimated from

the average velocity of each stream. The average velocity was obtained from the USGS

records for the stations along the stream. The uniformly distributed lateral inflow was

estimated by scaling the discharge records of the gaging station on a nearby watershed

with the fraction of the "unbalanced" drainage area to the area of the hydrologically

similar watershed. The "unbalanced" drainage areas are obtained from the calculations

shown in the spreadsheet in Table 5.

The "unbalanced" watershed areas that contribute directly to the distributed lateral

inflow are determined as the difference in the drainage area of a segment of the Illinois

River bounded by two adjacent gages and the total sum of the drainage areas of tributaries

that are discharging into the Illinois River between the two gages. For instance, the

difference in drainage areas above Kingston Mines and Havana gages is 2410 sq mi (Table

5). However, the sum of the watershed areas for Copperas Creek, Quiver Creek, and

Spoon River is 2024 sq mi. Therefore, the drainage area that is contributing to the

distributed inflow is 386 sq mi.

Big Bureau, Big, Spring, Hadley, and Bay Creeks were selected as hydrologically

similar streams for different segments of the lower Illinois River. The Big Bureau Creek

record was used for the distributed lateral inflow between Peoria L&D and Kingston

Mines, and for point inflow for Otter and Sugar Creeks. Big Creek data were used as

point lateral inflow for Copperas and Quiver Creeks, and for the distributed inflow

between Kingston Mines and Havana. Spring Creek records were used as point inflow for

Indian Creek and as distributed inflow between Havana and Meredosia and between

Ripley and the confluence of the La Moine River. Hadley Creek discharge records were

used for McKee, Mauvaise Terre, and Sandy Creeks as point inflow, and Bay Creek data

36

Table 5. Estimation of Drainage Area Ratio for Ungaged Watersheds

Illinois Illinois Drainage Representative Lag River River area stream for Drainage parameter

gaging station Tributaries Mile (sqmi) ungaged tributary area ratio (day)

Peoria L&D, TW 157.7 14,554 Mackinaw River 147.7 1,136 1

U D (Peoria-K.M.) --- 129 Big Bureau Cr. 0.66 1

Kingston Mines 144.4 15,819 Copperas Creek 137.4 127 Big Creek 3.15 1

Quiver Creek 122.6 261 Big Creek 6.48 1 Spoon River 120.5 1,855 1

U D (K.M.-Havana) --- 167 Big Creek 4.14 1

Havana 119.6 18,229 Otter Creek 111.8 126 Big Bureau Cr. 0.64 1 Sugar Creek 94.2 162 Big Bureau 0.83 1

Sangamon River 88.8 5,419 1 La Moine River 83.5 1,350 1

U D (Havana-La Grange) --- 362 Spring Creek 3.38 1

La Grange TW 80.2 25,648 Indian Creek 78.7 286 Spring Creek 2.67 1

U D (La Grange-Meredosia) --- 94 Spring Creek 0.88 1

Meredosia 71.3 26,028 McGee Creek 67.0 444 Hadley Creek 6.11 1

Mauvaise Terre Creek 63.2 178 Hadley Creek 2.44 1 Sandy Creek 50.0 166 Hadley Creek 2.28 1 Apple Creek 38.3 406 Bay Creek 10.3 1

Table 5. Estimation of Drainage Area Ratio for Ungaged Watersheds (Concluded)

Illinois Illinois Drainage Representative Lag River River area stream for Drainage parameter

gaging station Tributaries Mile (sqmi) ungaged tributary area ratio (day)

McCoupin Creek 23.2 961 1 Otter Creek 14.7 90 Bay Creek 2.28 1

U D (Meredosia-Grafton) --- 633 Bay Creek 16.04 1

Illinois River - 0.0 28,906 Mississippi River Confluence (0.2 mi

upstream of Grafton)

Notes: TW - Tailwater U D - Uniformly Distributed Lateral Inflow L&D - Lock and Dam

were applied to Apple and Otter Creeks. Hadley Creek were also used as distributed

inflow between Meredosia and Grafton.

Channel Roughness

The Manning roughness coefficients for the cross sections on the Illinois and

Sangamon Rivers were estimated from field reconnaissance surveys undertaken by the

USCOE. The roughness coefficients will be updated by calibrating the model to historical

flood data. In the UNET model, the Manning roughness coefficients are adjusted by

modifying the channel conveyance. The channel conveyance is the capacity of the channel

to transport water. Since the roughness coefficient is inversely proportional to the

conveyance, the roughness coefficient will increase as the conveyance is reduced and vice

versa. The model handles the change in conveyance by introducing a factor that is used to

multiply the conveyance. When the adjustment factor is less than unity, the conveyance is

reduced and the roughness coefficient is increased. If the factor is greater than one, the

conveyance is increased while the value of the roughness coefficient is reduced. A value

of one for the factor leaves both the conveyance and roughness coefficient unchanged.

The model provides two input options for modifying the channel conveyance. In

the first option, a pair of factors for the channel and overbank areas of the cross sections is

input into the model. Alternatively, if the roughness coefficient is observed to change with

stage and discharge, a table of discharge versus adjustment factor can be input to the

program.

Details of the procedure for calibration of the UNET model for the lower Illinois

River are presented in the next chapter.

Initial Flow Condition

The initial flow condition for each of the model reaches can be specified in the

time-series input file for the UNET model. The initial flow condition is specified in the

direction of the backwater flow, which is from downstream to upstream for subcritical

flow. If the initial condition is not specified, the model will assume steady-state subcritical

flow and generate initial discharge and water surface elevation for each of the model

39

reaches. No initial conditions were specified for the simulations presented in the next

chapter.

If storage areas are present in the system, the initial water surface elevation is

prescribed for each storage area. In the lower Illinois River, only one storage area was

included in the model. The storage volume and the initial water surface elevation were

included in the time-series data file.

40

MODEL CALIBRATION AND VERIFICATION

The UNET model can be applied to predict existing conditions, historical floods,

statistically significant flood events, and combinations of feasible scenarios. In order to

ensure that the simulation results are reliable and closely represent actual events, the

model has to be applied initially to simulate selected historical floods in the river system.

A level of error is pre-selected such that when the weighted sum of the differences

between computed and observed water surface elevations (WSEs) are below this level, the

computed WSE is taken as a true representation of the observed event. The fitting of the

computed WSE hydrographs to historical events is called model calibration. The WSE

and the flow computed by the model are adjusted to fit closely to observed data by

gradually varying the channel roughness coefficient until the difference between computed

and observed WSE is below the specified level of error. The Manning roughness

coefficient is expressed in the model in terms of the channel conveyance. Initial roughness

coefficient values are supplied to the model through the cross-section input file and are

then converted to conveyances. The conveyances are updated during the model

calibration by multiplying them by an adjustment factor that varies between 0 and 1. This

factor is included in the time-series input file and is varied until the error in the computed

WSE is below the tolerance.

Since subcritical flow is assumed in the UNET model, the calibration of the model

will start from the downstream end of the study reach and progress in the upstream

direction. The first calibrated section is between the downstream boundary and the next

upstream gage, the river section between Grafton and Hardin in the Peoria-Grafton

section of the Illinois River. If stage or water surface elevation is prescribed as the

downstream boundary condition (at Grafton), the conveyance in this river segment will be

adjusted until the computed stage or WSE hydrograph at Hardin matches the observed

record to within the error tolerance. The calibration then moves to the next section from

Hardin to Pearl. Since the WSE at Hardin has already been computed, it is necessary to

compute the WSE at Pearl. This will require a systematic adjustment of the conveyance

until the recorded WSE hydrograph at Pearl is closely matched. This procedure is

41

repeated for all river sections between adjacent upstream gages. Since there are 11 gages

on the lower Illinois River, ten roughness coefficients will be updated during the model

calibration. On the Sangamon River, the roughness coefficient for the section between

Oakford and the confluence of the Illinois River will be updated also.

Since the UNET model is based on an implicit solution procedure, the boundary

condition at the upstream end can be either WSE or flow. Because the USCOE gage at

Peoria has only stage records, WSE was prescribed at the upstream end of the study

reach. It was also used as the downstream boundary condition at Grafton. The April

1979 and March 1985 floods were selected for calibration and the data for four additional

events in 1973, 1974, 1982, and 1993 and were used to verify the calibrated parameters.

The results of these simulations are discussed in the following sections.

Calibration of 1979 and 1985 Floods

One of the flood events selected for calibration is the May 1979 flood ranked in

Table 6 as the fourth highest flow at Meredosia and the sixth highest flow at Kingston

Mines. The second event selected is the March 1985 flood ranked in Table 6 as second

and third at Meredosia and Kingston Mines, respectively. The flood of May 1943, the

highest flood at Meredosia and the second highest at Kingston Mines, was not selected

because of missing data in the records of some of the tributaries.

Figure 9 shows the April 1979 WSE hydrographs that were used as upstream

boundarycondition at Peoria L&D and downstream boundary condition at Grafton.

Figure 10 shows the computed WSE hydrographs obtained from the model calibration for

the nine gages between Peoria and Grafton. The computed hydrographs of surface water

elevation at Kingston Mines, Havana, and Beardstown [Figures 10(a) - 10(c)] fit the

observed records very closely. The differences between the computed and observed WSE

are generally less than 0.5 feet around the crest of the hydrographs, the only exception

being at Havana. The errors in the computed WSE are larger and generally above 1 foot

from La Grange to Grafton. The model underpredicted the WSE at these latter stations,

especially on the rising limb of the hydrograph. The computed hydrographs (Figure 10) fit

42

Table 6. Peak Floods in the Illinois River and Major Tributaries between Grafton and Peoria Lock and Dam

Ill. River Ill. River Kingston

Item Macoupin Meredosia LaMoine Sangamon Spoon Mines Mackinaw

Rank 1 1 18 1 28 2 11 Qp 40,000 123,000 14,500 123,000 12,900 83,100 18,200

Date 5/18/43 5/26/43 5/21/43 5/20/43 5/20/43 5/23/43 5/19/43

Rank 10 2 1 21 3 3 10 QP 19,400 122,000 28,800 26,500 29,200 78,800 18,400

Date 2/24/85 3/10/85 3/7/85 2/25/85 3/6/85 3/6/85 2/25/85

Rank 3 3 6 2 9 1 1 Qp 26,700 112,000 21,000 68,700 21,000 88,800 51,000

Date 12/4/82 12/12/82 12/5/82 12/5/82 4/4/83 12/7/82 12/6/82

Rank 2 4 34 3 30 6 6 Qp 27,800 111,000 8,120 55,900 12,600 72,300 23,600

Date 4/12/79 4/19/79 4/12/79 4/15/79 3/31/79 3/24/79 3/5/79

Rank 22 5 29 6 1 7 30 QP 12,800 110,000 10,000 42,900 36,400 71,900 7,910

Date 1/21/74 6/29/74 6/2/74 6/25/74 6/24/74 5/25/74 6/5/74

Rank 31 6 28 29 17 4 8 QP 9,140 104,000 10,700 23,700 16,900 77,200 20,000

Date 2/21/82 3/24/82 3/16/82 3/20/82 7/21/82 3/22/82 3/12/82

Rank 20 7 22 4 21 14 4 QP 13,300 102,000 12,400 45,800 16,000 63,300 29,700

Date 4/23/73 5/2/73 4/22/73 4/25/73 4/24/73 4/25/73 4/21/73

Figure 9. Water surface elevation hydrographs used as (a) upstream boundary condition at Peoria Lock and Dam and (b) downstream boundary condition at Grafton

44

Figure 10. Computed and observed water surface elevation hydrographs at (a) Kingston Mines, (b) Havana, (c) Beardstown, (d) La Grange Lock and Dam, (3) Meredosia,

(f) Valley City, (g) Florence, (h) Pearl, and (I) Hardin

45

Figure 10 (continued)

46

Figure 10. (continued).

47

Figure 10. (continued)

48


49

the observed WSE much better on the falling limb. Peak WSEs are also underpredicted, but the underestimations are within 1 foot at these downstream gages except Pearl.

Since the model predictions degenerate from La Grange L&D downstream, it seems that the UNET model cannot effectively simulate the operations of the lock and dam and the flow through the structure. This apparent problem could be avoided by setting up the study reach as two separate subreaches: the La Grange Pool and the Alton Pool. The subdivision of the lower Illinois River into two reaches implies that La Grange L&D will serve as the downstream boundary for the La Grange Pool as well as the upstream boundary for the Alton Pool. Figure 11 shows the April 1979 WSE hydrograph that was used as the boundary condition at La Grange, and Figure 12 shows the plots of the station hydrographs resulting from the two-reach model calibrations. Table 7 shows the coefficient of determination values, and Figure 13 shows the computed peak WSE for the two-reach model and the computed results for the one-reach simulations and the observed peak WSEs at the 11 stations for comparison. It should be noted that the peak WSE profile is not a snapshot of the water surface elevation at a particular instance in time. It is, however, a plot of the peaks of the WSE hydrographs at each of the cross sections. The WSE profile for the two-reach model (Figure 13) more closely fit the observed elevations than the one-reach profile. The two computed profiles are relatively indistinguishable from Peoria to Havana but differ downstream from Havana, the greatest difference occurring between Beardstown and Meredosia. The maximum difference in either of the computed profiles and the observed peak WSE occurred around Meredosia.

Because of the improvement in the computation results for the two-reach model, the model calibration for the March 1985 flood was also carried out for both one-reach and two-reach models. The coefficient of determination for the stations is shown in the second column of Table 7, the computed WSE profiles are shown in Figure 14 and the station hydrographs are shown in Figure 15. Figure 14 also shows the recorded peak WSE at the gages. Both of the computed profiles in the figure seemed to fit the observed data very well. The one-reach peak WSE profile provides a better fit to the observed data while the two-reach simulations slightly over-predicted the peak WSE at the stations. The differences between the two simulated profiles, and the differences between either profile

50

Figure 11. Water surface elevation hydrograph used as boundary condition at La Grange Lock and Dam for the two-reach model

51

Figure 12. Water surface elevation hydrographs computed (two-reach model) and observed at (a) Kingston Mines, (b) Havana, (c) Beardstown, (d) Meredosia, (e) Valley City,

(f) Florence, (g) Pearl, and (h) Hardin.

52

Figure 12 (continued)

53


54


55

Figure 13. Comparison of peak water surface elevation profiles for one-reach and two-reach models with observed peak stages for the 1979 flood at gaging stations along the lower Elinois River.

56

Table 7. Coefficient of Determination Values for the Observed and Computed Stage Hydrographs Using Two-Reach Model

Calibration Verification Station 1979 1985 1982 1974 1973

Peoria L&D TW Kingston Mines 0.9979 0.9941 0.9978 0.9990 0.9975

Havana 0.9904 0.9820 0.9970 0.9899 0.9928 Beardstown 0.9618 0.9662 0.9888 0.9469 0.9967

La Grange L&D TW Meredosia 0.9991 0.9990 0.9963 0.9987 0.9973 Valley City 0.9960 0.9921 0.9768 0.9871 0.9833 Florence 0.9821 0.9872 0.9614 0.9792 0.9669

Pearl 0.9808 0.9759 0.9069 0.9560 0.9572 Hardin 0.9785 0.9612 0.8813 0.9204 0.9660 Grafton

Number of Days 135 151 45 107 151

Figure 14. Comparison of peak water surface elevation profiles for one-reach and two-reach models with observed peak stages for the 1985 flood at gaging stations along the lower Illinois River

58



59


60


61


62

and the observed peak WSE are less than 0.5 feet. A comparison of the 1979 and 1985

coefficient of determination values in Table 7 indicates that the two-reach model

calibrations for the 1979 flood generally provided better representation of observed WSE

hydrographs than the 1985 simulations. However, by examining the computed peak WSE

profile for 1985 (see Figure 14), the two-reach calibration for the 1985 event is shown to

be better than the 1979 calibration (Figure 13). As a result of these observations, the

Manning roughness coefficients, obtained from the two-reach calibration for both floods

events, were averaged to obtain the roughness coefficient values that will be used in all

subsequent simulations. Table 8 lists these roughness coefficient values. The roughness

coefficient is 0.02 for the channel areas of all cross sections. The value of the roughness

coefficients for the overbank areas varies between 0.03 and 0.1. A value of 1.0 has been

used for the areas behind the levees to prevent any flow in this region.

Model Verification with Other Floods

Using the roughness coefficient values obtained from the calibration of the 1979

and 1985 floods, the model was applied to simulate the December 1982, June 1974, April

1973, and July 1993 flood events. These events are ranked as third, fifth, seventh, and

twelfth at Meredosia and as first, seventh, fourteenth, and eighteenth at Kingston Mines,

respectively. The computed two-reach peak WSE profiles and the corresponding WSE

hydrographs at the gages are plotted in Figures 16-21. Figures 16, 18, and 20 are profiles

for the 1982, 1974, and 1973 events, respectively, and Figures 17, 19, and 21 show the

corresponding WSE hydrographs at the gage locations. Figure 22 shows the peak WSE

profile for the 1993 flood.

The computed two-reach peak WSE profiles for the 1982 flood, shown in Figure

16, fit the observed peaks better than the one-reach results. The computed profiles are

similar except in the Alton Pool, most especially from Hardin to Grafton,where they differ

by more than 1 foot. The WSE hydrographs for the two-reach simulations are plotted in

Figure 17. The computed and observed hydrographs at Kingston Mines, Beardstown, and

Meredosia matched the observed data very closely. Although the errors in the

computed peak WSE at Valley City, Florence, and Pearl are within 0.5 feet, the predicted

63

Table 8. Manning Roughness Coefficients for Reaches along the Illinois River

Upstream-downstream Roughness coefficient River reach River miles Channel Left bank Right bank

Peoria L&D - Kingston Mines 157.8-144.4 0.02 0.10 0.10 Kingston Mines - Havana 144.4-119.6 0.02 0.10 0.03-0.10 Havana - Beardstown 119.6-88.8 0.02 0.10 0.03-0.10 Beardstown - La Grange L&D 88.8-80.2 0.02 0.03-0.1 0.10-1.00 La Grange L&D - Meredosia 80.2-71.3 0.02 0.03-0.1 0.10-1.00 Meredossia - Valley City 71.3-61.3 0.02 0.03-0.1 0.10-1.00 Valley City - Florence 61.3-56.0 0.02 0.03-0.1 0.10-1.00 Florence - Pearl 56.0-43.2 0.02 0.03-0.1 0.10-1.00 Pearl-Hardin 43.2-21.5 0.02 0.03-0.1 0.10-1.00 Hardin - Grafton 21.5-0.0 0.02 0.03-0.1 0.10-1.00


65



66


67


68


69


70


(f) Florence, (g) Pearl, and (h) Hardin

71


72


73


74


75



76


77


78


79

Figure 22. Comparison of peak water surface elevation profiles for two-reach model with observed peak stages for the 1993 flood at gaging stations along the lower Illinois River

80

hydrographs still lagged the observed hydrographs by one to two days. The model also

underpredicted the peak WSE at Hardin by about 0.5 feet, and the computed hydrograph

has a broader crest in comparison to the observed record. The tail of the hydrograph is

grossly overpredicted at this station by as much as 4 feet. The poor performance of the

model for this flood may be due to the ice conditions on the Illinois River in December or

the variation in magnitude and timing of the flow from Macoupin Creek.

Figures 18 and 19 show the simulation results for the June 1974 flood. In Figure

18, the accuracy of the model is observed to deteriorate downstream of La Grange L&D

for the one-reach model, but the two-reach simulation results closely match the observed

WSE at each of the stations. Figure 19 shows the two-reach WSE hydrographs at the

stations. The computed hydrographs closely matched the observed hydrographs. The

computed hydrographs, however, lagged the observed data by up to one day from Valley

City to Hardin. The greatest difference in WSE is at Hardin, an observation also made in

previous simulations.

The one-reach peak WSE for April 1973 is very poor in comparison to the two-

reach simulations shown in Figure 20. The two-reach model significantly under-estimated

the peak WSE downstream of Meredosia, but the errors are less than 0.25 feet at these

stations. Figure 21 shows the computed and observed hydrographs for the nine interior

stations for the two-reach model. The hydrographs are generally in phase at each station,

but a difference of up to 0.75 feet was observed at Florence, Pearl, and Hardin.

The flood peak of 51,200 cfs, recorded at Kingston Mines on July 29, 1993, has a

return interval of approximately three years for a 53-year (1941-1993) annual peak flow

record. A flood peak of 89,630 cfs was observed at Meredosia on August 1, 1993, and

this discharge has an estimated return period of about five years for the same length of

record. The low return periods of the flood peaks at these two stations indicate that the

peak stages on the lower Illinois River in July and August of 1993 were not produced by a

major flood on the river. Instead, the prolonged record stages in the Mississippi River

raised the stages along the Illinois River to levels much higher than the corresponding 3-

to 5-year return period flows.

81

The extent of the influence of the backwater effect can be determined by examining

the peak stages at Illinois River stations upstream of Grafton. The peak stage at Grafton

was 441.8 feet-msl, which is the highest in the 53-year record. The peak stage at

Meredosia, which is 70.8 miles upstream, was 3.10 feet higher than the peak stage at

Grafton. This difference in WSEs between the two stations is the lowest when compared

with the March 1985, April 1979, and December 1982 floods where the differences in

WSE were at least four times higher (15.13, 13.93, and 12.80 feet, respectively). It is,

therefore, apparent that the backwater effect extended up to Meredosia and probably as

far upstream as Kingston Mines.

By ranking the annual peak WSE (1941-1993) for the stations between Meredosia

and Kingston Mines, the ranks for the 1993 flood stages are:

Peak WSE

Station River Mile (ft-msl) Date Rank

Grafton 0.0 441.8 8/1/93 1

Meredosia 70.8 444.9 7/28/93 4

La Grange 80.2 445.95 7/27/93 5

Beardstown 88.1 446.6 7/28/93 6

Havana 119.6 447.9 7/29/93 7

Kingston Mines 145.6 448.4 7/30/93 14

The rank at Meredosia is 4 and the rank increased by one for consecutive upstream

stations up to Havana. At Kingston Mines the rank for the peak stage jumped to 14.

From the listing in the table above, there is an increase in rank of 6 between Grafton and

Havana (119.6 miles) and an increase of 7 between Havana and Kingston Mines (26.0

miles). The sharp jump in rank over the short distance between Havana and Kingston

Mines indicates that the backwater effect may not have reached as far as Kingston Mines.

The rank of 14 at Kingston Mines corresponds to about a four-year reccurrence interval,

which is about the same as for the peak flood at that station. The influence of the rising

82

stages on the Mississippi seemed to end somewhere between Havana (R.M. 119.6) and

Kingston Mines (R.M. 145.6).

Since the backwater effect resulting from the Mississippi River flood extended

upstream of Havana, the UNET model validation of the 1993 flood was carried out for

both the Alton and La Grange Pools. Figure 22 shows the computed peak WSE profile

and the observed data. The figure shows that peak WSEs were somewhat underpredicted

at all stations in the Alton Pool except Valley City. However, the predicted WSEs are all

within 0.5 feet of the observed peaks.

83

PEAK STAGE REDUCTION FROM LEVEE STORAGE OPTIONS

This chapter outlines the next phase of the Managed Flood Storage Options

project. This phase is currently being implemented and will be reported in the third project

report. Current tasks include the determination of flood profiles for 10-, 25-, 50-, and

100-year return periods; examination of the variability of flow in the Illinois River and

flood peak timing from Illinois River tributaries; and evaluation of flood reduction due to

conversion of some of the levees and drainage districts in the Alton and La Grange Pools

to the managed flood storage option.

Recurrence Interval Flood Profiles

As stated in previous chapters, the simulation of flood elevation profiles along the

lower Illinois River requires the selection of flood hydrographs for the tributaries and

stage hydrographs for the upstream and downstream boundaries on the main river stem.

For instance, the selection of the appropriate tributary flow hydrographs that will generate

the 100-year peak WSE profile along the Illinois River is a major task in the modeling

process. This is because various combinations of tributary flows can, theoretically, be

applied to generate the 100-year flood elevation profile in different reaches of the river.

The problem becomes more complex because a 100-year flood elevation profile is

generally not generated by a 100-year flood flow in the Illinois River. There is no direct

relationship between stage and discharge, so the 100-year flood elevation may be

generated by more than one flow at each gaging station. This has already been observed

in some of the historical floods on the Illinois River. For instance, in December 1982, the

third highest flood of 112,000 cubic feet per second or cfs occurred at Meredosia and the

highest flood of 88,000 cfs at Kingston Mines in the concurrent record from 1941-1993.

In April 1979, a flood of 111,000 cfs occurred at Meredosia and 72,000 cfs at Kingston

Mines ranked fourth and sixth at these stations, respectively. It would be expected that

the 1982 WSE profile will be higher than the 1979 profile. However, from the plot of the

peak WSE at the gaging stations along the lower Illinois River (Figure 23), the 1979

84

Figure 23. Observed peak water surface elevation profiles for the April 1979 and December 1982 floods on the lower Illinois River

85

WSEs are higher than the 1982 elevations. The approach adopted in this project is

described in the next few paragraphs.

Using the flood frequency analysis described in the first project report (Singh,

1996), the 2-, 10-, 25-, 50-, 100-, and 500-year floods were evaluated for the five stations

that represent the ungaged tributaries (Big Bureau, Big, Spring, Hadley, and Bay Creeks).

Table 9 shows the results of these analyses. In addition, the table contains the frequency

analysis results for the five gaged tributaries (Mackinaw, Spoon, Sangamon, and La Moine

Rivers, and Macoupin Creek) on the lower Illinois River.

The first six top floods at each of the ten stations were selected and the recorded

discharges for ten days preceding and for ten days succeeding the occurrence of the peak

discharge were plotted for each flood. Each of the six hydrographs was first normalized

with the corresponding peak discharge and then plotted on the same 20-day base with the

peaks occurring on the same day as shown in Figures 24-33. A representative normalized

hydrograph for a station was then obtained from the six hydrographs. As much as

possible, the representative hydrographs were drawn as closely as possible to the first

three top-flood hydrographs. The synthesized station hydrographs that will be used in the

simulation for a return period were obtained by multiplying the ordinates of the

representative normalized hydrograph by the corresponding peak flow from Table 9. The

synthesized hydrographs for the various return periods were read into the Data Storage

System database and will be for each of the tributaries.

The next step was to provide the appropriate time lag for each of the synthetic

hydrographs by examining the timing of the peak floods for the flood events in 1943,

1974, 1979, 1982, and 1985 used for model calibration and validation. The time lag of the

peaks of the recorded hydrographs at the tributary gages relative to Meredosia (Table 10)

seem to be between six and seven days. Peoria L&D seems to be ahead of Meredosia by

about five days, suggesting that the upstream boundary hydrograph at Peoria L&D will lag

behind the tributary station hydrographs by one to two days. One of the series of

simulations that will be carried out will involve the variation of the lags so that the flood

peak from some of the tributaries coincides with the flood wave on the Illinois River. This

86

Table 9. Tributary Flood Peaks (cfs) at Various Recurrence Intervals

Tributary 2-year 10-year 25-year 50-year 100-year 500-year

Mackinaw River 10,300 27,600 39,600 50,000 62,000 97,800 Spoon River 14,000 26,100 32,300 37,800 41,000 51,800 Sangamon River 24,500 46,300 59,000 72,000 87,500 138,000 LaMoine River 12,000 23,000 26,800 31,000 35,000 45,000 Macoupin Creek 10,800 22,800 29,300 34,600 39,600 59,400 Big Bureau Creek 4,210 8,100 10,000 11,500 12,800 15,500 Big Creek 773 1,157 1,336 1,465 1,592 1,882 Spring Creek 1,840 5,750 8,300 10,500 13,200 20,000 Hadley Creek 7,339 12,650 15,240 17,337 19,811 25,162 Bay Creek 5,134 10,541 13,210 15,291 17,491 23,499

Figure 24. Normalization of the hydrographs of the top six floods, recorded at the Mackinaw River gage near Green Valley, by their respective peak flows and selection of a representative discharge hydrograph

88

Figure 25. Normalization of the hydrographs of the top six floods, recorded at the Spoon River gage at Seville, by their respective peak flows and selection of a representative discharge hydrograph

89

Figure 26. Normalization of the hydrographs of the top six floods, recorded at the Sangamon River gage near Oakford, by their respective peak flows and selection of a representative discharge hydrograph

90

Figure 27. Normalization of the hydrographs of the top six floods, recorded at the La Moine River gage at Ripley, by their respective peak flows and selection of a representative discharge hydrograph.

91

Figure 28. Normalization of the hydrographs of the top six floods, recorded at the Macoupin Creek gage near Kane, by their respective peak flows and selection of a representative discharge hydrograph

92

Figure 29. Normalization of the hydrographs of the top six floods, recorded at the Big Bureau Creek gage at Princeton, by their respective peak flows and selection of a representative discharge hydrograph

93

Figure 30. Normalization of the hydrographs of the top six floods, recorded at the Big Creek gage near Bryant, by their respective peak flows and selection of a representative discharge hydrograph

94

Figure 31. Normalization of the hydrographs of the top six floods, recorded at the Spring Creek gage at Springfield, by their respective peak flows and selection of a representative discharge hydrograph.

95

Figure 32. Normalization of the hydrographs of the top six floods, recorded at the Hadley Creek at Kinderhook gage, by their respective peak flows and selection of a representative discharge hydrograph

96

Figure 33. Normalization of the hydrographs of the top six floods, recorded at the Bay Creek at Pittsfield gage, by their respective peak flows and selection of a representative

discharge hydrograph.

97

Table 10. Days Flood Peak at Meredosia Lags behind Those at Other Stations

Macoupin Meredosia Sangamon La Moine Spoon Mackinaw Kingston Mines Year Qp Lag Qp Qp Lag Qp Lag Qp Lag Qp Lag Qp Lag

1943 40,000 -8 123,000 123,000 7 14,500 6 12,900 6 18,200 8 83,100 4 1985 19,400 - 122,000 26,500 5 23,000 3 29,200 4 18,400 - 78,800 4 1982 26,700 -8 112,000 68,700 7 21,000 6 21,000 7 51,000 7 88,800 5 1979 27,800 -7 111,000 55,900 4 3,120 7 12,600 6 23,400 - 72,300 4 1974 12,800 - 110,000 42,900 4 10,000 3 36,400 5 7,910 - 71,900 5

Representative (-7) to 5-6 5-6 5-6 7-8 4 Value of Lag (-8)

Note: Data have been taken from Table 25 of Singh (1996). Qp = Observed flood peak, cfs Estimated lag for Peoria L&D is 5 days

type of analysis will be useful for predicting the highest floods that are feasible in the

Illinois River.

The first set of simulation runs will involve the prediction of the 100-year peak

WSE. Since the recurrence interval peak elevations for the stations have already been

determined from the frequency analysis of stage data as shown in Table 11 (Singh, 1996),

the objective is to compute water elevations that will match the 100-year elevations at the

stations by adjusting the lateral inflows from the tributaries. First, the normalized WSE

hydrographs at Peoria L&D, La Grange L&D and Grafton were determined (see Figures

34-36) and then multiplied by their respective peak WSE for the 100-year recurrence

interval. These synthesized hydrographs will be used as upstream and downstream

boundary conditions in the following simulations. Starting from Peoria, the next step is to

match the 100-year peak WSE at Kingston Mines, the next station downstream of Peoria.

This will be accomplished by starting with the 100-year discharge value for the gage on

the Mackinaw River near Green Valley, adjusting this value gradually and using it after

each adjustment to multiply the normalized hydrograph at the station until the computed

peak WSE at Kingston Mines matched the 100-year water elevation. The goal in this

effort is to compute a peak WSE that closely approximates the 100-year elevation. By so

doing, it will be shown that the shape and duration of the computed hydrograph may not

be a true representation of the actual 100-year hydrograph.

The final adjusted peak discharge for Green Valley will be checked with Table 9 to

determine the recurrence interval of the Mackinaw River flow that contributed to the 100-

year WSE in the Peoria-Kingston Mines reach of the Illinois River. The above procedure

will be repeated for the remaining nine reaches between the gages downstream of

Kingston Mines. As mentioned previously, the recurrence interval for the tributary flows

will not necessarily be 100 years. The 10-, 25-, and 50-year water elevation profiles along

the lower Illinois River will also be reproduced in a similar manner.

Effect of Changing Discharge and Variable Tributary Flows

The objective of these simulations will be to develop all feasible WSE profiles in

the Illinois River from a combination of tributary and lateral inflows of various recurrence

99

Table 11. Peak Stages (cfs) for Gaging Stations in the Lower Illinois River at Various Recurrence Intervals

Gaging Station 2-year 10-year 25-year 50-year 100-year 500-year

Peoria L&DTW 447.02 452.30 454.43 455.92 457.28 460.21 Kingston Mines 445.68 450.82 453.02 454.55 456.00 459.19 Havana 442.49 447.79 450.11 451.68 453.14 456.26 Beardstown 439.08 445.51 447.91 449.52 451.02 454.21 La Grange L&D TW 437.62 443.73 446.20 447.90 449.48 452.92 Meredosia 436.72 442.92 445.38 447.08 448.59 451.89 Valley City 434.85 441.22 443.84 445.61 447.29 450.88 Florence 433.73 440.19 442.85 444.66 446.35 449.97 Pearl 431.18 437.44 440.15 442.07 443.91 448.01 Hardin 427.58 434.55 437.49 439.53 441.47 445.91 Grafton 425.23 432.52 435.81 438.09 440.24 444.86

Figure 34. Normalization of the stage hydrographs of the top six floods, recorded at the Illinois River gage at Peoria Lock and Dam Tailwater, by their respective peak stages

and selectionof a representative stage hydrograph

101

Figure 35. Normalization of the stage hydrographs of the top six floods, recorded at the Illinois River gage at La Grange Lock and Dam Tailwater, by their respective peak stages

and selection of a representative stage hydrograph

102

Figure 36. Normalization of the stage hydrographs of the top six floods, recorded at the Mississippi River gage at Grafton which is near the confluence of the Illinois River, by their respective peak stages

and selection of a representative stage hydrograph

103

intervals. The analysis will also include the use of various time lags in such a way that the

arrival of peak flow at the mouth of some of the tributaries coincide with the peak flood

on the Illinois River. This flood coincidence analysis will be very useful in defining the

feasibility of floods of higher magnitudes than has been observed to date and the

probability of their occurrences. The recurrence probability of these higher floods will be

extrapolated by plotting the return frequency floods on the same graph as the computed

floods.

Examination of the flood stage profiles will help to determine the peak WSEs

generated by the 10-, 25-, 50-, and 100-year flood stages at Kingston Mines and

Meredosia.

Effect of Levee Flood Storage on Flood Peaks

The study of the reduction of peak flood stages due to flood storage in selected

levees will be carried out mainly in the Alton Pool since the levees in this section of the

Illinois River have been subjected to more frequent overtopping than the levees in the La

Grange Pool because of the backwater effects from the Mississippi River. The simulations

will examine existing conditions as well as other feasible situations.

Existing conditions will be examined in three ways. One series of simulation runs

will be based on the assumption of a wall at the crown of the levees so that no levee

overtopping will occur. Another set of runs will allow levee overtopping with provision

for possibility of levee failures. In addition, the much often discussed situation of no

levees on the Illinois River will be simulated by changing the cross-sectional elevations in

the data input file. These simulations will be run for 10-, 25, 50-, and 100-year flows

developed from the analyses in the previous section.

By selecting individual and combinations of levees in the Alton and La Grange

Pools and creating openings 500-1000 feet wide and 4-6 feet deep along the crown of

some of the levees, simulation runs for 10-, 25-, 50-, and 100-year flow will be performed.

The results of these simulations in comparison to the previous simulation runs will be used

to determine the reductions in peak flood that can be attained due to the conversion of

104

some of the levees to temporary storage areas and to evaluate the duration of holding of

water in these storage basins.

105

SUMMARY

The unsteady flow model, UNET, has been tested and validated for the lower

Illinois River reach between Peoria L&D and Grafton. The model was applied to simulate

the May 1979 and March 1985 floods, which are ranked, respectively, fourth and second

at Meredosia, and sixth and third at Kingston Mines. The computed water surface

elevations were compared with the recorded events at eight stage-gaging stations between

Peoria L&D and Grafton. The computed WSE hydrographs for the 1979 flood event fit

the observed data generally within 0.5 feet for the stations in the La Grange Pool.

However, the model errors were generally over one foot at the stations in the Alton Pool.

The deterioration in the accuracy of the one-reach model (Peoria L&D to Grafton) for the

Alton Pool led to the development of a two-reach model (Peoria L&D to La Grange L&D

and La Grange L&D to Grafton). The computed two-reach WSE for the 1979 flood fit

the observed WSEs more closely than the one-reach model.

The 1985 flood was also simulated with the one-reach and the two-reach models,

but the one-reach model results were better than the two-reach model results. However,

the differences between observed and computed peak WSE for the two models were less

than 0.5 feet. The two-reach model calibrations for the 1979 flood provided better

approximation of the observed peaks than the 1985 simulations. However, the stage

hydrographs were better predicted for the 1985 flood than for the 1979 flood. Because of

the mixed results, the roughness coefficients adopted for subsequent analyses were

obtained by averaging the roughness coefficients for the 1979 and 1985 calibrations.

Further testing of the UNET model for the lower Illinois River included the

simulation of the December 1982, June 1974, April 1973, and July 1993 flood events.

The two-reach model WSE profiles for these floods fit the observed data much better than

the one-reach model profiles. The two-reach model will, therefore, be adopted for all

future simulations.

The predicted WSE for the 1982 flood event were all within 0.5 feet of the

observed data at all the gaging stations. The predicted hydrographs lagged the observed

hydrographs by one to two days at the stations downstream of Meredosia. At Hardin, the

106

crest of the predicted WSE hydrograph was broader than the observed hydrograph. The

differences in the computed and observed WSE hydrographs may be due to errors caused

by the variation in the timing of the flood flow from Macoupin Creek.

The computed WSE hydrographs for the June 1974 flood event closely matched

the observed hydrographs at all the stations, but there were lags of up to one day in the

peak WSE hydrographs from Valley City to Hardin. The model underestimated the

observed peak WSE for the 1973 flood by less than 0.25 feet at the stations. The peaks of

the WSE hydrographs were generally in phase with the observed hydrographs, but there

were differences of up to 0.75 feet in the peak WSE between Florence and Hardin.

The 1993 peak discharges at Meredosia and Kingston Mines were estimated to

have return periods of approximately three to five years. The low return periods at these

stations implied that the high stages in the Illinois River in 1993 were not due to a major

flood on the river. Instead, the prolonged and unprecedented flooding on the Mississippi

River, which produced the highest historical stage at Grafton, caused backwater effects

that extended upstream on the Illinois River. The backwater effects were shown to affect

stages at Havana (R.M. 119.6), but it did not seem to affect stages at Kingston Mines

(R.M. 145.6). The prediction of the 1993 flood event was shown to be accurate within

0.5 feet of the observed peak WSE.

Work in Progress

The model simulations, for the analysis of flood-stage reduction due to levee

storage options, required the development of typical discharge hydrographs for each of the

tributary stations and stage hydrographs at Peoria L&D, La Grange L&D, and Grafton.

They also required the estimation of appropriate time lags for the tributary station

hydrographs. This information will be used in the development of the 100-year stage

profile along the Illinois River and for the generation of all feasible flood stages and their

probability of occurrence.

The ordinates of the six top floods at the gages on the five major tributaries

(Mackinaw, Spoon, Sangamon, and La Moine Rivers, and Macoupin Creek) and at the

five stations representing the ungaged tributaries (Big Bureau, Big, Spring, Hadley, and

107

Bay Creeks) were normalized with the peak discharges of corresponding hydrographs.

Representative normalized flood hydrographs were constructed for each station and then

converted to actual flood hydrographs by multiplying the ordinates by the corresponding

10-, 25-, 50-, and 100-year discharges for each station. These synthetic hydrographs will

be used in the simulation of WSE profiles for the various return periods. Appropriate time

lags for the synthetic hydrographs have been determined from the timing of the flood

peaks for the 1943, 1974, 1979, 1982, and 1985 flood events. The lags for the gaged

tributaries were estimated as approximately between six and seven days relative to

Meredosia.

The first series of simulations that is currently underway includes the prediction of

the 100-year WSE profiles for the La Grange and Alton Pools. Starting from Peoria, the

tributary flows will be adjusted until the peak WSE at the Illinois River gages closely

match the 100-year flood stages derived from the frequency analyses of the gage records.

The next task will be to develop of all feasible WSE profiles obtained by combining

tributary and lateral inflows of various recurrence intervals and adjusting the time lags of

the discharge hydrographs so that the timing of some of the tributary floods coincides with

those on the Illinois River.

The final stage of the simulations will be the prediction of peak stage reductions

for levee storage options mainly in the La Grange Pool and some in the Alton Pool. The

model simulations will include scenarios of 1) no levee failure by assuming a vertical wall

along the crowns of the levees to determine maximum flood stages, 2) the flood stages

along the Illinois River if the levees had not been constructed, and 3) the restricted flood

stages for safety of levees with some of them converted to the levee storage option.

108

REFERENCES

Akanbi, A.A., and N.D. Katopodes. 1988. A Model for Flood Propagation on Initially

Dry Land, Journal of Hydraulic Engineering, ASCE, Vol. 114, No. 7, pp. 689-706.

Bathe, K., and E.L. Wilson. 1976. Numerical Methods in Finite Element Analysis,

Prentice-Hall, Inc., Englewood Cliffs, NJ.

Boss International, 1993. FastTABS Hydrodynamic Modeling Reference Manual,

Version 1, Madison, WI.

Franz, D.D. 1990. Notes for Short Course and Input Description for FEQ Version 6.0.

Fread, D., 1985. "Channel Routing" in Hydrological Forecasting, M.G. Anderson and

T.P. Burt (eds.), John Wiley & Sons, New York, NY, pp. 437-503.

Froehlich, D.C., 1989. FESWMS-2DH Finite-Element Surface Water Modeling System;

Two-Dimensional Flow in a Horizontal Plane: User Manual, U.S. Department of

Transportation, Federal Highway Administration, Publication No. FHWA-RD-88-

177.

Healy, R.W. 1979. River Mileages and Drainage Areas for Illinois Streams —Volume 2,

Illinois River Basin, U.S. Geological Survey Water Resources Investigations, #79-

111, Champaign, JX, 302p.

HEC, 1993. UNET — One-dimensional Unsteady Flow through a Full Network of Open

Channels; User's Manual, U.S. Army Corps of Engineers, Hydraulic Engineering

Center, Davis, CA, May, 240p.

Preissmann, A., 1960. Propagation of Translatory Waves in Channels and Rivers, First

Congres de 1'Assoc. Francaise de Calcul, Grenoble, France, pp. 433-442.

109

Singh, K.P. 1996. Managed Flood Storage Option for Selected Levees along the Lower

Illinois River for Enhancing Flood Protection, Agriculture, Wetlands, and

Recreation, First Report: Stage and Flood Frequencies and the Mississippi

Backwater Effects, Contract Report 590, Illinois State Water Survey, Champaign, EL,

138p.

U.S. Army Corps of Engineers (USCOE), 1996. Upper Mississippi River Basin, Illinois

Waterway Nine-foot Channel, Master Water Control Manual, Appendix 7: La

Grange Lock & Dam, U.S. Army Corps of Engineers, North Central Division, Rock

Island, EL.

USCOE, 1987. Upper Mississippi River Basin, Illinois Waterway Nine-Foot Channel,

Master Reservoir Regulation Manual, U.S. Army Engineer District, Rock Island,

Corps of Engineers, Rock Island, IL.

110

Documents

Managed flood storage option for selected levees along the lower